Latest News
This week, Nathan, Zahra, and Ainaz delivered a workshop entitled *Beating an AI Wizard: Can You Get AI to Tell You the Secret Password? for Oshawa Public Libraries.
The aim of this workshop was to defeat an AI Wizard who holds a series of passwords. Before tackling the grand challenge, participants first explored concepts of Artificial Intellgience, Large Language Models, Guardrails and Prompt Engineering. They were introduced to a range of case-studies that highlight the gaps in security, and completed an activity to foster their jailbreaking skills, all in effort of bolstering their chances of taking down Gandalf!
This weekend, Joelma travelled to Montreal to present research entitled “Who Benefits from AI Explanations? Towards Accessible and Interpretable Systems”. Joelma presented this work at the Workshop on Explainable Artificial Intelligence (XAI), hosted at IJCAI 2025, the 34th International Joint Conference on Artificial Intelligence.
As more people and organizations turn to artificial intelligence (AI) for ideas and answers, it is essential to question whther the information AI provides is reliable and accurate.
In this interview, Peter answers questions on trust, uncertainty, and self-awareness, describing some of the ongoing projects we have in our lab, and ultimately cautioning people to be `rational skeptics’; to take advantage of AI but to not be naïve to its challenges and downsides.
Peter and Steve were both invited to attend the sixth Trusting Intelligent Machines workshop, hosted at Schloss Rauischholzhausen in Germany. The workshop included experts on trust, explainability, and artificial intelligence from academia and instrustry across Germany, the UK, Finland, and Canada. The theme of the workshop was exploring the personal and social consequences of the widespread outsourcing of human cognition to AI systems.
This week Peter and Stavros traveled to Detroit for the 24th International Conference on Autonomous Agents and Multiagent Systems (AAMAS).
There, Peter presented co-authored work with Zahra, entitled Evaluating Prompt Engineering Techniques for Accuracy and Confidence Elicitation in Medical LLMs at the 7th International Workshop on EXplainable, Trustworthy, and Responsible AI and Multi-Agent Systems (EXTRAAMAS).
